qiqihaer / RandLA-Net-pytorch

RandLA-Net's implementation with Pytorch

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Random sampling or fixed sampling?

minwang-ai opened this issue · comments

Hi,

According to the paper:
Random Sampling (RS): We implement random sam-
pling with the python numpy package. Specifically, we
first use the numpy function numpy.random.choice() to
generate K indices. We then gather the corresponding
spatial coordinates and per-point features from point
clouds by using these indices.

However, the sub_points and pool_index are generated by dividing sampling ratio and there is no random sampling there.

Could you tell me how did you implement random sampling?